Symptoms of anxiety and depression and risk of atrial fibrillation-the HUNT study

(1)

Symptoms of anxiety and depression and risk of atrial ﬁ brillation — The HUNT study

Tingting Feng

^a,

⁎ , Vegard Malmo

^b,c

, Lars E. Laugsand

^b,d

, Linn B. Strand

^a

, Lise T. Gustad

^b,e

, Hanne Ellekjær

^f,g

, Jan P. Loennechen

^b,c

, Kenneth Mukamal

^h

, Imre Janszky

^a,i,j

aDepartment of Public Health and Nursing, Norwegian University of Science and Technology, Trondheim, Norway

bDepartment of Circulation and Medical Imaging, Norwegian University of Science and Technology, Trondheim, Norway

cClinic of Cardiology, St. Olavs Hospital, Trondheim, Norway

dDepartment of Emergency Medicine, St. Olavs Hospital, Trondheim, Norway

eDepartment of Medicine, Levanger Hospital, Nord-Trøndelag Hospital Trust, Levanger, Norway

fStroke Unit, Department of Internal Medicine, St Olav's Hospital, Trondheim, Norway

gDepartment of Neuromedicine and Movement Science, Norwegian University of Science and Technology, Trondheim, Norway

hDepartment of Medicine, Beth Israel Deaconess Medical Center, Boston, MA, USA

iDepartment of Neurology, Medical School, University of Pécs, Pécs, Hungary

jInstitute of Behavioural Sciences, Semmelweis University, Budapest, Hungary

a b s t r a c t a r t i c l e i n f o

Article history:

Received 6 September 2019

Received in revised form 6 November 2019 Accepted 13 November 2019

Available online 14 November 2019 Keywords:

Atrialﬁbrillation Depression Anxiety

Background:Atrialﬁbrillation (AF) is the most common sustained cardiac arrhythmia. Anxiety and depression may activate the autonomic nervous system which is likely to play an important role in the etiology of AF. How- ever, little is known about the association between symptoms of anxiety and depression and risk of AF.

Objective:This study aimed to assess the association between symptoms of anxiety and depression and risk of AF.

Methods:In a population-based study, 37,402 adult residents were followed for incident AF from 2006 to 2008 until 2015. Participants were classiﬁed according to data on anxiety and depression symptoms. Cox proportional regression models were used to adjust for common AF risk factors.

Results:During a median follow-up of 8.1 years, 1433 (3.8%) participants developed AF. In comparisons with no anxiety symptoms, the multivariable-adjusted hazard ratios (HRs) were 1.1 (95% CI: 0.9–1.5) for mild to moderate anxiety symptoms and 1.0 (95% CI: 0.8–1.4) for severe anxiety symptoms. In comparisons with no depression symptoms, the multivariable-adjusted HRs were 1.5 (95% CI: 1.2–1.8) for mild to moderate depression symptoms and 0.9 (95% CI: 0.6–1.3) for severe depression symptoms. Recurrent anxiety/depression symptoms were not associated with increased AF risk.

Conclusions:In this large, population-based study, we found no evidence of an association between symptoms of anxiety or severe depression and AF risk, even for recurrent anxiety or depression symptoms. An unexpected association of symptoms of mild to moderate depression with increased AF risk requires conﬁrmation in other studies. Ourﬁndings add to the sparse literature on symptoms of anxiety and depression and risk of AF.

creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction

Atrialﬁbrillation (AF) is the most common sustained cardiac arrhythmia and is associated with increased risk of mortality and stroke as well as increased health care costs [1]. To be able to better prevent and treat AF, it is important to identify its modiﬁable risk factors [2].

Symptoms of depression and anxiety have been relatively strongly

⁎ Corresponding author at: Department of Public Health and General Practice, Faculty of Medicine, Norwegian University of Science and Technology, NO-7489 Trondheim, Norway.

E-mail address:tingting.feng@ntnu.no(T. Feng).

https://doi.org/10.1016/j.ijcard.2019.11.107

Contents lists available atScienceDirect

International Journal of Cardiology

j o u r n a l h o m e p a g e :w w w . e l s e v i e r . c o m / l o c a t e / i j c a r d

(2)

associated with cardiac diseases, including coronary heart disease [3], acute myocardial infarction [4] and heart failure [5]. Furthermore, depression and anxiety may activate the autonomic nervous system [6]

which is likely to play an important role in the etiology of AF [7]. How- ever, only limited research has explored the effect of psychological distress on AF onset [8,9]. Therefore, the aim of this large population-based study was to assess the prospective association of symptoms of anxiety and depression with AF risk.

2. Methods 2.1. Study population

All 93,860 residents≥20 years of age in Nord-Trøndelag County in Norway were invited to participate in the third Nord-Trøndelag Health (HUNT 3) study from October 2006 to June 2008. Of these, 50,804 participants (54%) answered questionnaires and underwent a clinical examination. Holmen et al. have described the HUNT study in more detail [10].

We excluded 1598 participants from the analysis who had a history of AF at baseline and 11,804 participants with missing response for depression or anxiety symptoms, although we tested the inclusion of participants with missing data in imputation analyses described below.

Details about inclusions are provided in eFig. 1.

2.2. Measures of depression and anxiety symptoms

The Norwegian version of the Hospital Anxiety and Depression Scale (HADS) was used to assess symptoms of anxiety (HADS-A) and depression (HADS-D) during the previous week [11]. The HADS is a valid and reliable instrument across various patient samples and settings [12]. It comprises 14 self-rated items, with seven items forming the Anxiety subscale (HADS-A) and seven items forming the Depression subscale (HADS-D) [13]. The HADS-A reﬂects symptoms of worry and tension, while HADS-D reﬂects symptoms of anhedonia and loss of interest.

The items on the HADS-A subscale are:“(1) I feel tense or wound up;

(2) I get a sort of frightened feeling as if something bad is about to hap- pen; (3) Worrying thoughts go through my mind; (4) I can sit at ease and feel relaxed; (5) I get a sort of frightened feeling like butterﬂies in the stomach; (6) I feel restless and have to be on the move; (7) I get sud- den feelings of panic.”The items on the HADS-D subscale are:“(1) I still enjoy the things I used to enjoy; (2) I can laugh and see the funny side of things; (3) I feel cheerful; (4) I feel as if I am slowed down; (5) I have lost interest in my appearance; (6) I look forward with enjoyment to things; (7) I can enjoy a good book or radio or TV program.”Each item on the questionnaire is scored from 0 (no symptom) to 3 (highest symptom level). Thus, participants can score between 0 and 21 for either symptoms of anxiety or depression. Participants were included in the subsequent analyses only if they answered all of the 14 items. The two subscales were divided into 3 categories: (1) having no symptoms of depression or anxiety (scoreb8), (2) having mild to moderate symptoms (score 8–10), and (3) having severe symptoms (score≥11), respectively [14].

Among the 37,402 eligible participants in HUNT 3, 24,706 individuals had available information on HADS from the second wave of the HUNT Study (HUNT 2) conducted between 1995 and 1997 [10]. A combined burden of anxiety symptoms in HUNT 2 and 3 was categorized into 3 groups: (1) no anxiety symptoms (scoringb11 on HADS-A in HUNT 2 and 3), (2) anxiety symptoms at one time (scoring≥11 on HADS-A in one of the HUNT studies), and (3) anxiety symptoms at both times (scoring≥11 on HADS-A in both HUNT studies). A combined burden of depression symptoms in HUNT 2 and 3 was categorized into 3 groups: (1) no depression symptoms (scoringb11 on HADS-D in HUNT 2 and 3), (2) depression symptoms at one time (scoring≥11 on HADS-D in one of the HUNT studies), and (3) depression symptoms at both times (scoring≥11 on HADS-D in both HUNT studies).

2.3. Atrialﬁbrillation

From the baseline examination until November 30, 2015, AF diagnoses were retrieved from discharge registers at two hospitals that to- gether serve the entire population of Nord-Trøndelag County. We used code I48 from the International Classification of Diseases Tenth Re- vision to screen for patients with possible AF. Medical records of these patients were then reviewed by experts, and AF was adjudicated based on electrocardiographic criteria recommended by the European Society of Cardiology [15]. Persons who only had an episode of AF within thefirst 7 days after cardiac surgery, during the acute phase of a myocardial infarction or during episodes of hemodynamic instability (e.g., sepsis or non-cardiac surgery) were not regarded as having incident AF. If information from medical records was insufficient for exact classification of the diagnosis, two physicians (HE and JPL) evaluated the available information separately. Only cases where both physicians concurred were regarded as AF. The rest were classified as possible AF and were not regarded as AF cases in the main analyses. The validation process is described in more detail elsewhere [16].

2.4. Covariates

A clinical examination was conducted by trained nurses. Height and weight were measured barefoot and wearing light clothing; height was measured to the nearest cm and weight to the nearest 0.5 kg. Body mass index was calculated as body weight in kilograms divided by the squared value of height in meters. Non-fasting blood samples were an- alyzed for glucose, triglycerides, high-density lipoprotein cholesterol (HDL-C) and high-sensitivity C-reactive protein (CRP) [10]. Self- reported data obtained at the baseline included smoking status (never, former, or current), alcohol consumption (abstainers, light drinkers, moderate drinkers, or heavy drinkers), physical activity (inac- tivity, moderate activity or high activity), occupation (desk work, light industry work, technicians or heavy physical work), and marital status (cohabitation or no cohabitation). Information on common chronic disorders was self-reported and included: angina pectoris, stroke, asthma, osteoarthritis, kidney disease, hyperthyroidism, rheumatoid arthritis, sarcoidosis, ankylosing spondylitis, cancer, epilepsy, chronic bronchitis, emphysema or COPD, psoriasis, diabetes, sleep apnea, acute myocardial infarction, and heart failure. Apart from self-report, information on history of acute myocardial infarction and heart failure was also retrieved from hospital registers [17].

2.5. Statistical analyses

Baseline characteristics were presented as mean ± standard devia- tion for continuous variables and percentages for categorical variables.

Cox proportional regression models were used to assess the association of symptoms of depression and anxiety with subsequent risk of AF. Par- ticipants who reported no symptoms (i.e., scoringb8 on HADS-A or HADS-D) were regarded as the reference group. Time was deﬁned as days from inclusion to either incident AF or censoring due to death from other causes (n = 2457), emigration from the county (n = 82) or end of follow-up. We calculated hazard ratios (HRs) and 95% conﬁ- dence intervals (CI). We included age and sex in model 1; in addition, weight, height, smoking, occupation type, marital status, physical activity, and alcohol consumption were added as potential confounders in model 2. In further analyses, we also adjusted for chronic disorders, i.e., all diseases mentioned in the previous section were included in the regression models as part of a composite variable having the value of 1 in the presence of any of the condition and 0 otherwise (model 3) and for metabolic components (i.e., blood glucose, blood pressure, triglycerides, HDL-C and CRP) (model 4). We tested linear as well as quadratic trends for the associations of symptoms of anxiety and depression with AF risk. To explore dose-dependent associations, we presented the hazard ratios of AF in the form of smoothed curves where

(3)

symptoms of anxiety and depression were used as continuous variables [18].

In separate analyses, we assessed the relative risk of AF according to episodes of anxiety and depression symptoms in HUNT 2 and 3. Partic- ipants without symptoms in any of the HUNT studies were regarded as the reference group.

The proportional hazards assumption was tested by comparing−ln- ln survival curves and by performing tests on Schoenfeld residuals for each covariate. We found no violations of the proportionality assumption.

To assess effect modiﬁcation, we conducted analyses stratiﬁed by age, sex, and chronic diseases.

In sensitivity analyses, we regarded possible or single-episode AF during follow-up as events. Furthermore, to address the possibility of reverse causation as an explanation for the observed associations, we excluded theﬁrst 2 years of follow-up and repeated the analyses. We also used multiple imputation withﬁve imputations (mi command in Stata) to examine whether the complete-case approach yielded biased estimates [19]. Lastly, to assess to what extent symptoms of anxiety/depression may improve the ability of a multivariable model to predict a new-onset AF, we developed logistic regression-based multivariable prediction models with and without symptoms of anxiety/depression.

We calculated the area under the ROC (receiver operating characteris- tic) curve (AUC) in these models [20]. All statistical analyses were conducted using Stata/MP 15.1 for Windows (College Station, TX: StataCorp LLC).

3. Results

3.1. Baseline characteristics

Characteristics of participants with and without subsequent AF are presented in Table 1. During a median follow-up of 8.1 years (288,460 person-years), 1433 (3.8%) participants developed AF. Using a cutoff score of 11 and above for severe symptoms, 2.2% of the participants reported symptoms of depression and 4.9% reported symptoms of anxiety. The participants that developed AF were older and more likely

to be men, inactive, heavy drinkers, and score higher on the HADS-D subscale.

3.2. Relative risk of atrialﬁbrillation in relation to symptoms of anxiety and depression

Table 2shows HRs for incident AF according to symptoms of depression and anxiety in HUNT 3. In comparisons with no symptoms of anxiety, the multivariable-adjusted HRs (model 2) were 1.1 (95% CI:

0.9–1.5) for symptoms of mild to moderate anxiety and 1.0 (95% CI:

0.8–1.4) for symptoms of severe anxiety, respectively. In comparisons with no symptoms of depression, the multivariable-adjusted HRs were 1.5 (95% CI: 1.2–1.8) for symptoms of mild to moderate depression and 0.9 (95% CI: 0.6–1.3) for symptoms of severe depression, respectively (P= 0.002 for quadratic trend). Additional adjustment for chronic disorders or metabolic status did not materially change the estimates.

Fig. 1presents the relative risks for atrialﬁbrillation in the form of a smoothed curve when symptoms of anxiety and depression were used as continuous variables. There was no clear sign of an increased risk for with symptoms of severe depression or anxiety in these analyses.

Table 3shows the HRs for AF according to episodes of symptoms of anxiety and depression in HUNT 2 and 3. Participants with recurrent symptoms of anxiety or depression did not have higher relative risks of incident AF compared to those without symptoms of anxiety or depression at any of the HUNT studies.

3.3. Effect modiﬁcation

Overall, we found limited evidence of major effect modiﬁcation by age, sex, or chronic disease (eTables 1–3). In general, the HRs tended to be higher among younger individuals, men, and participants without chronic disease.

3.4. Sensitivity analyses

We documented 1165 AF cases after the second year of follow-up.

Exclusion of theﬁrst 2 years of follow-up had little effect on the results (eTable 4). The results were also consistent with the main analyses when possible or single-episode AF events (n = 99) were regarded as AF during follow-up. Finally, when we performed multiple imputation, the results were similar to those in the primary analysis (eTable 5).

3.5. AF identiﬁcation ability of prediction models with and without symptoms of anxiety and depression

The AUC in the multiadjusted logistic regression model without symptoms of anxiety and depression was 0.8504, while AUC in the logistic regression model with symptoms of anxiety and depression was 0.8504 and 0.8506, respectively. Thus, adding symptoms of anxiety/depression to the model did not improve its ability to distinguish between patients with AF and no AF.

4. Discussion

In this large, population-based study, we found no evidence of an association between severe symptoms of depression and AF risk, even when symptoms of depression were recurrent. We also found no association of symptoms of anxiety with AF risk. Unexpectedly, only symptoms of mild to moderate depression were associated with an increased AF risk.

Only four previous studies have assessed the prospective association of symptoms of anxiety or depression with AF risk, and all had potential limitations in their assessments of exposure or outcome, and none had repeated measures of symptoms of anxiety or depression. In the Fra- mingham Offspring study which had a relatively small sample size Table 1

Baseline characteristics of participants in the total population according to atrialﬁbrilla- tion vs. no atrialﬁbrillation during follow-up.

Variable No. of

subjects Total population

AF during follow-up

No AF during follow-up

Total % (n) 37,402 % 3.8

(1433) 96.2 (35,969) Variables, % (n)

Female 21,122 56.5 41.5

(595) 57.1 (20,527)

Diabetes mellitus 1636 4.4 10.3

(148)

4.1 (1488)

Current smoker 6081 16.7 12.5

(173)

16.8 (5908)

Heavy drinkers 523 1.4 2.5 (35) 1.4 (488)

Technicians 7234 20.2 12.3

(164)

20.5 (7070)

Physically inactive 7470 20.3 24.1

(336)

20.1 (7134)

Cohabitation 22,798 61.0 64.7

(927) 60.9 (21,871) Mean (SD)

Age, years 37,402 53.4

(15.2) 70.1 (10.9)

52.8 (15.0) Body mass index, kg/m² 37,289 27.2 (4.4) 28.2 (4.6) 27.1 (4.4) HADS—depression score (HUNT

3)

37,402 3.3 (2.9) 4.0 (2.9) 3.3 (2.9) HADS—anxiety score (HUNT 3) 37,402 4.0 (3.3) 3.7 (3.0) 4.0 (3.3) AF: atrialﬁbrillation; HADS: Hospital Anxiety and Depression Scale.

(4)

(n = 3682) and a 10-year follow-up [9], symptoms of anxiety in men (HR: 1.1; 95% CI, 1.0–1.3) had a weak association with AF risk while there was no such an association in women (HR: 1.0; 95% CI, 0.8–1.3).

In the Multi-Ethnic Study of Atherosclerosis (n = 6644) [21], symptoms of depression were associated with higher AF risk (HR: 1.3; 95% CI, 1.0–1.7). However, few details of these results are available, since these were published only as an abstract. Furthermore, both the Fra- mingham Offspring study and the Multi-Ethnic Study of Atherosclerosis used instruments (i.e., the Tension and Symptoms of Anxiety Scales [9]

and Depression Scale of Epidemiologic Studies [22], respectively) that include somatic symptoms. Consequently, the potential overlap of somatic symptoms caused by physical illness with that of psychological distress limited the ability of these studies to examine the genuine effects of core psychological and cognitive symptoms of anxiety and depression on AF risk. On the other hand, our study used the HADS scale, which replaces somatic symptoms with non-somatic alternatives [11].

Thus, in our study, we were able to examine the association of core psychological and cognitive symptoms of anxiety and depression with AF risk. In the Women's Health Study [8], symptoms of depression was unrelated to AF risk (HR: 1.0; 95% CI, 0.8–1.3). The Mental Health Inventory-5, which has similar features as HADS in terms of exclusion of somatic symptoms, was used to assess symptoms of depression [8].

However, AF events were self-reported in this study, which would tend to lower the specificity of the AF diagnoses. A Danish matched cohort study compared AF risk in all Danes initiating antidepressant medication (n = 785,254) with that in a 1:5-matched sample from the general population [23]. The study defined depression as the condition within the month before initiation of antidepressant medication. Sub- stantially increased AF risk was observed even before antidepressant medication (HR = 3.18; 95% CI: 2.98–3.39) and within thefirst month (4.29; 95% CI: 3.94–4.67) after antidepressant initiation, but AF risk de- creased 6–12 months after antidepressant initiation (HR = 1.11; 95% CI:

1.06–1.16). However, antidepressant medication is only a proxy for depression and these medications have common indications beyond depression as well [24], such as pain or insomnia. For example, a study conducted infive European countries showed that between 1.1 and 7.4% of prescribed antidepressants were used for sleeping disorders [24]. Another European study found that 25% of prescribed antidepressants were used for chronic pain [25]. Furthermore, the study retrieved AF diagnosis from registers without further manual verification, which could have resulted in misclassification.

Our study revealed an unexpected inverted U-shaped association between symptoms of depression and AF risk, which is inconsistent with the limited previous research. The underlying mechanism is un- clear. Depression may lead to increased inﬂammation [26], oxidative stress [27] and sympathetic activation which in turn could increase the risk for AF. However, depression has also been associated with para- sympathetic suppression [28], which might actually reduce AF risk, thus potentially explaining the null effects of symptoms of severe depression.

It is also possible that the null effect for symptoms of anxiety and severe depression may be related to cardiac effects of different antidepressant medications. Selective serotonin reuptake inhibitors (SSRIs) seem to have a cardio-protective proﬁle [29]. SSRIs have been documented to improve glucose metabolism, dyslipidemia, and reduce inﬂammatory markers, which may contribute to reduced AF risk [30]. In the Table 2

Hazard ratios (95% conﬁdence intervals) for atrialﬁbrillation during follow up according to symptoms of anxiety (HADS-A) and depression (HADS-D) in HUNT 3.

Model 1 Model 2 Model 3 Model 4

Events/person-years 1433/288,460 1235/266,671 1186/256,765 1205/260,753

HADS-A

0–7 Reference Reference Reference Reference

8–10 1.0 (0.8–1.3) 1.1 (0.9–1.5) 1.1 (0.8–1.4) 1.1 (0.8–1.4)

≥11 1.1 (0.8–1.4) 1.0 (0.8–1.4) 1.0 (0.7–1.4) 1.1 (0.8–1.5)

Pfor linear trend 0.5092 0.7865 0.9603 0.6695

Pfor quadratic trend 0.8193 0.5454 0.5736 0.7501

HADS-D

0–7 Reference Reference Reference Reference

8–10 1.3 (1.0–1.6) 1.5 (1.2–1.8) 1.4 (1.1–1.7) 1.5 (1.2–1.8)

≥11 0.9 (0.7–1.3) 0.9 (0.6–1.3) 0.8 (0.6–1.2) 0.8 (0.6–1.2)

HADS-D and HADS-A score:b8: no symptoms of depression or anxiety, 8–10: mild to moderate symptoms,≥11: severe symptoms.

Model 1: adjusted for age and sex.

Model 2: Model 1 + weight, height, smoking status, occupation, marital status, physical activity and alcohol consumption.

Model 3: Model 2 + chronic disorders.

Model 4: Model 2 + metabolic components (i.e., blood glucose, blood pressure, triglycerides, high-density lipoproteins and C-reactive protein.).

Fig. 1.Smoothed curve for hazard ratios of atrialﬁbrillation and symptoms of anxiety and depression at baseline (adjusted for sex, age, height, smoking status, level of education, marital status, physical activity and alcohol consumption).

(5)

aforementioned Danish study, AF risk was high 30 to 15 days before antidepressant medication initiation and after theﬁrst month of medication [23]. The association gradually attenuated over the fol- lowing 6–12 months (HR = 1.11; 95% CI: 1.06–1.16). The attenua- tion of the association could reﬂect the possibility that depression- related AF may be a time-dependent event and that AF risk might decrease after an initial higher-risk period [23]. In our study, we did not have information about when depression may have started nor whether our participants were medicated, which limited our ability to explore the reason for the null association between symptoms of depression and AF risk.

Additionally, one can speculate that the higher AF risk among those with symptoms of mild/moderate depression–when compared to those without symptoms of depression and to those with symptoms of severe depression–might be due to differential help-seeking behav- iors and differential detection rate of AF. Also, a prior study examining the effects of depression and antidepressant therapy on the risk of mortality and cardiovascular diseases showed that individuals with the most severe depression beneﬁted most from antidepressant therapy re- garding mortality and cardiovascular diseases, and those who had only mild depression showed no cardiovascular risk reduction [31]. Thus, individuals with symptoms of severe depression in our study may have adhered more regularly to antidepressant treatment compared to those with symptoms of mild to moderate depression, which might explain the observed null association between symptoms of severe depression and AF risk.

4.1. Strengths and limitations

The strengths of our study derive from the population-based design, the homogeneity of the population (eliminating confounding by racial differences in AF risk) [32], its high stability (b0.3% net migration/

year), the uniformly organized health care system which was equally available for every citizen and the relatively high response rate. Com- prehensive data including those on a wide range of chronic disorders allowed us to thoroughly characterize participants and minimize confounding. We had data on repeated assessments of symptoms of anxiety and depression, which offered a unique opportunity to examine how AF risk was related to changes in symptoms of anxiety or depression over time. Furthermore, careful veriﬁcation of AF diagnoses ensured the min- imization of the misclassiﬁcation of endpoints.

Some limitations warrant discussion. First, we relied on a self- reported scale to measure symptoms of anxiety and depression and lacked a clinician-administered interview to establish psychiatric diagnoses. While the HADS has been documented to be a valid and reliable instrument for screening among a wide variety of patient samples and

the general population, its ability to differentiate between anxiety and depression and its capability as a case-ﬁnder (i.e., to identify cases of clinical depression/anxiety) are imperfect [33,34]. This may have led to some misclassiﬁcation of symptoms of depression and anxiety. Coro- nary heart disease is generally more strongly associated with anxiety and depression when these conditions are clinically diagnosed than when self-reported questionnaires alone are used [35]. Thus, the association between anxiety/depression and AF risk might be stronger if clinical interviews were available. Second, we lacked data on antidepressant medication. Third, the average HADS scores on symptoms of anxiety were generally lower compared to other studies in Norway and other European countries [33,36–38]. It might be possible that individuals with substantial anxiety symptoms were less prone to participate in the study. Fourth, we did not have updated information of symptoms of anxiety and depression during follow-up after baseline.

This could lead to measurement error if participants changed their psychological status during follow-up. Because information on symptoms of anxiety and depression was gathered before diagnosis of AF, any misclassification would be expected to be unrelated to the outcome, i.e., non-differential misclassification [39]. Because non-differential misclassification tends to lead to weakening/dilution of the measure of association, i.e., a “bias towards the null”, it might explain the null association observed between symptoms of severe depression and AF risk. However, we had two measurements on HADS at or before baseline, and even recurrent symptoms of depression or anxiety were not associated with AF risk. In a previous study conducted in the same population, symptoms of depression and anxiety as measured by HADS were prospectively associated with both myocardial infarction and heart failure [4,5]. Lastly, AF can be occult and we only identified AF cases that came to clinical attention. Therefore, under-detection of AF, particularly when paroxysmal, is likely to exist in ourfindings. However, the consequence of under-detection, namely lower sensitivity, is generally less threatening to the validity of an epidemiological study than is lower specificity [16,39]. Nonetheless, this may have precluded us from identifying a true association between depression or anxiety and risk of AF if one exists.

5. Conclusion

In this large, population-based study, we found no evidence of an association between symptoms of anxiety or severe depression and AF risk, even for recurrent symptoms of anxiety or depression.

An unexpected association of symptoms of mild to moderate depression with increased AF risk requires conﬁrmation in other studies.

Table 3

Episodes of symptoms of anxiety and depression in HUNT 2 and 3, and risk for atrialﬁbrillation.

Model 1 Model 2 Model 3 Model 4

Episodes of anxiety

No anxiety Reference Reference Reference Reference

Anxiety at one time 1.2 (0.9–1.6) 1.1 (0.9–1.5) 1.1 (0.8–1.4) 1.2 (0.9–1.6)

Anxiety at two times 0.9 (0.5–1.6) 1.0 (0.5–1.9) 0.9 (0.5–1.8) 1.0 (0.5–2.0)

Episodes of depression

No depression Reference Reference Reference Reference

Depression at one time 1.2 (0.9–1.6) 1.2 (0.9–1.6) 1.1 (0.8–1.6) 1.2 (0.9–1.6)

Depression at two times 0.9 (0.4–1.9) 0.9 (0.4–2.0) 0.9 (0.4–2.1) 0.8 (0.3–1.9)

Model 1: adjusted for age and sex.

Model 2: Model 1 + weight, height, smoking status, occupation, marital status, physical activity and alcohol consumption.

Model 3: Model 2 + chronic disorders.

Model 4: Model 2 + metabolic components (i.e., blood glucose, blood pressure, triglycerides, high-density lipoproteins and C-reactive protein).

(6)

Declaration of competing interest

None of the authors have any conﬂicts of interest.

Acknowledgements

The HUNT study is a collaboration between the Faculty of Medicine and Health Sciences, Norwegian University of Science and Technology, the Nord-Trøndelag County Council and the Norwegian Institute of Pub- lic Health. All laboratory analyses were performed andﬁnanced by the Health Trust of Nord-Trøndelag. We thank the Department for Research and Development, and clinicians at the Medical Department, Nord- Trøndelag Hospital Trust, Norway, for extracting the data from the patient registers.

Funding

Grants are from Norwegian Heart and Lung Association and Liaison Committee for Education, Research and Innovation in Central Norway.

The funding sources had no role in the study design, in the collection, analysis and interpretation of data, in the writing of the report, or in the decision to submit the article for publication.

Appendix A. Supplementary data

Supplementary data to this article can be found online athttps://doi.

org/10.1016/j.ijcard.2019.11.107.

References

[1] P.A. Wolf, J.B. Mitchell, C.S. Baker, W.B. Kannel, R.B. D’Agostino, Impact of atrialﬁbril- lation on mortality, stroke, and medical costs, Arch. Intern. Med. 158 (1998) 229–234.

[2] D.H. Lau, S. Nattel, J.M. Kalman, P. Sanders, Modiﬁable risk factors and atrialﬁbrilla- tion, Circulation 136 (2017) 583–596.

[3] R.M. Carney, K.E. Freedland, Depression and coronary heart disease, Nat. Rev.

Cardiol. 14 (2017) 145.

[4] L.T. Gustad, L.E. Laugsand, I. Janszky, H. Dalen, O. Bjerkeset, Symptoms of anxiety and depression and risk of acute myocardial infarction: the HUNT 2 study, Eur. Heart J.

35 (2013) 1394–1403.

[5] L.T. Gustad, L.E. Laugsand, I. Janszky, H. Dalen, O. Bjerkeset, Symptoms of anxiety and depression and risk of heart failure: the HUNT study, Eur. J. Heart Fail. 16 (2014) 861–870.

[6] S.D. Kreibig, Autonomic nervous system activity in emotion: a review, Biol. Psychol.

84 (2010) 394–421.

[7] P.-S. Chen, L.S. Chen, M.C. Fishbein, S.-F. Lin, S. Nattel, Role of the autonomic nervous system in atrialﬁbrillation: pathophysiology and therapy, Circ. Res. 114 (2014) 1500–1515.

[8] W. Whang, K.W. Davidson, D. Conen, U.B. Tedrow, B.M. Everett, C.M. Albert, Global psychological distress and risk of atrialﬁbrillation among women: the Women’s Health Study, J. Am. Heart Assoc. 1 (2012), e001107.

[9] E.D. Eaker, L.M. Sullivan, M. Kelly-Hayes, R.B. D’Agostino Sr., E.J. Benjamin, Tension and anxiety and the prediction of the 10-year incidence of coronary heart disease, atrialﬁbrillation, and total mortality: the Framingham Offspring Study, Psychosom.

Med. 67 (2005) 692–696.

[10] J. Holmen, K. Midthjell, Ø. Krüger, A. Langhammer, T.L. Holmen, G.H. Bratberg, et al., The Nord-Trøndelag Health Study 1995–97 (HUNT 2): objectives, contents, methods and participation, Norsk Epidemiologi. 13 (2003) 19–32.

[11] A. Mykletun, E. Stordal, A.A. Dahl, Hospital Anxiety and Depression (HAD) scale: factor structure, item analyses and internal consistency in a large population, BrJPsychiatry. 179 (2001) 540–544.

[12] I. Bjelland, A.A. Dahl, T.T. Haug, D. Neckelmann, The validity of the Hospital Anxiety and Depression Scale: an updated literature review, J. Psychosom. Res. 52 (2002) 69–77.

[13]M.A. Annunziata, B. Muzzatti, G. Altoe, Deﬁning hospital anxiety and depression scale (HADS) structure by conﬁrmatory factor analysis: a contribution to validation for oncological settings, Ann. Oncol. 22 (2011) 2330–2333.

[14] A. Zigmond, R. Snaith, The hospital anxiety and depression scale, ActaPsychiatrScand. 67 (1983) 361–370.

[15] A.J. Camm, P. Kirchhof, G.Y. Lip, U. Schotten, I. Savelieva, S. Ernst, et al., Guide- lines for the management of atrialﬁbrillation, Eur Heart J 31 (19) (2010) 2369–2429.

[16]V. Malmo, A. Langhammer, K.H. Bønaa, J.P. Loennechen, H. Ellekjaer, Validation of self-reported and hospital-diagnosed atrialﬁbrillation: the HUNT study, Clin Epidemiol. 8 (2016) 185.

[17] B. Morkedal, L.J. Vatten, P.R. Romundstad, L.E. Laugsand, I. Janszky, Risk of myocardial infarction and heart failure among metabolically healthy but obese individuals:

HUNT (Nord-Trondelag Health Study), Norway, J Am Coll Cardiol. 63 (2014) 1071–1078.

[18] L.A. Sleeper, D.P. Harrington, Regression splines in the Cox model with application to covariate effects in liver disease, J. Am. Stat. Assoc. 85 (1990) 941–949.

[19] D.B. Rubin, Multiple imputation for nonresponse in surveys, Wiley Series in Proba- bility and Statistics, , 1987.

[20] M.J. Pencina, R.B. D’Agostino Sr., R.B. D’Agostino Jr., R.S. Vasan, Evaluating the added predictive ability of a new marker: from area under the ROC curve to reclassiﬁcation and beyond, Stat. Med. 27 (2008) 157–172.

[21]P.K. Garg, W.T. O’Neal, A.V.D. Roux, A. Alonso, E. Soliman, S. Heckbert, Depressive symptoms and risk of incident atrialﬁbrillation: the multi-ethnic study of atherosclerosis, Circulation 137 (2018) AMP60.

[22] R.N. Carleton, M.A. Thibodeau, M.J. Teale, P.G. Welch, M.P. Abrams, T. Robinson, et al., The center for epidemiologic studies depression scale: a review with a theoretical and empirical examination of item content and factor structure, PLoS One 8 (2013), e58067.

[23] M. Fenger-Grøn, M. Vestergaard, H.S. Pedersen, L. Frost, E.T. Parner, A.R. Ribe, et al., Depression, antidepressants, and the risk of non-valvular atrialﬁbrillation: a nation- wide Danish matched cohort study, Eur. J. Prev. Cardiol. 26 (2) (2018) 187–1952047487318811184.

[24] V. Abbing-Karahagopian, C. Huerta, P. Souverein, F. De Abajo, H. Leufkens, J. Slattery, et al., Antidepressant prescribing inﬁve European countries: application of common deﬁnitions to assess the prevalence, clinical observations, and methodological impli- cations, Eur J Clin Pharmacol. 70 (2014) 849–857.

[25] Verhaak PF, de Beurs D, Spreeuwenberg P. What proportion of initially prescribed antidepressants is still being prescribed chronically after 5 years in general practice?

A longitudinal cohort analysis. BMJ open. 2019;9:e024051.

[26] G.E. Miller, E. Blackwell, Turning up the heat: inﬂammation as a mechanism linking chronic stress, depression, and heart disease, Curr. Dir. Psychol. Sci. 15 (2006) 269–272.

[27] P. Palta, L.J. Samuel, E.R. Miller III, S.L. Szanton, Depression and oxidative stress: results from a meta-analysis of observational studies, Psychosom. Med. 76 (2014) 12.

[28]J.M. Gorman, R.P. Sloan, Heart rate variability in depressive and anxiety disorders, Am. Heart J. 140 (2000) S77–S83.

[29] M.H. Nezafati, M. Vojdanparast, P. Nezafati, Antidepressants and cardiovascular ad- verse events: a narrative review, ARYA Atherosclerosis. 11 (2015) 295.

[30] M.H. Nezafati, A. Eshraghi, M. Vojdanparast, S. Abtahi, P. Nezafati, Selective serotonin reuptake inhibitors and cardiovascular events: a systematic review. Journal of research in medical sciences: the ofﬁcial journal of Isfahan University of Medical, Sciences 21 (2016).

[31] H.T. May, T. Bair, B. Reiss-Brennan, S. Knight, J. Anderson, B. Horne, et al., The association of antidepressant and statin use to future death and incident cardiovascular disease varies by depression severity, J. Am. Coll. Cardiol. 65 (2015) A1443.

[32] L. Mou, F.L. Norby, L.Y. Chen, W.T. O’Neal, T.T. Lewis, L.R. Loehr, et al., Lifetime risk of atrialﬁbrillation by race and socioeconomic status: ARIC Study (Athero- sclerosis Risk in Communities), Circ. Arrhythm. Electrophysiol. 11 (2018), e006350.

[33] M.W. Nortvedt, T. Riise, B. Sanne, Are men more depressed than women in Norway?

Validity of the Hospital Anxiety and Depression Scale, J. Psychosom. Res. 60 (2006) 195–198.

[34] A.J. Mitchell, N. Meader, P. Symonds, Diagnostic validity of the Hospital Anxiety and Depression Scale (HADS) in cancer and palliative settings: a meta-analysis, J. Affect.

Disord. 126 (2010) 335–348.

[35] I. Janszky, A. Ahlbom, J. Hallqvist, S. Ahnve, Hospitalization for depression is associated with an increased risk for myocardial infarction not explained by lifestyle, lipids, coagulation, and inﬂammation: the SHEEP Study, Biol. Psychiatry 62 (2007) 25–32.

[36] A. Jörngården, L. Wettergen, L. von Essen, Measuring health-related quality of life in adolescents and young adults: Swedish normative data for the SF-36 and the HADS, and the inﬂuence of age, gender, and method of administration, Health and Quality of Life Outcomes 4 (2006) 91.

[37] W. Häuser, K.-H. Janke, B. Klump, A. Hinz, Anxiety and depression in patients with inﬂammatory bowel disease: comparisons with chronic liver disease patients and the general population, Inﬂamm. Bowel Dis. 17 (2010) 621–632.

[38]C. Bocéréan, E. Dupret, A validation study of the Hospital Anxiety and Depression Scale (HADS) in a large sample of French employees, BMC Psychiatry. 14 (2014) 354.

[39] K.J. Rothman, S. Greenland, T. Lash, Modern Epidemiology, 1, Lippincott Wil- liams & Wilkins Foundation for Alcohol Research & Education Level, Philadel- phia, 2008 40.